ViDi Labs. SD/HD CCTV Test Chart. v.4.x. Instructions for setup and usage. Designed and Produced by. ViDi Labs Pty.Ltd. 2010~2015 ABN

Transcription

1 ViDi Labs SD/HD CCTV Test Chart v.4.x Instructions for setup and usage Designed and Produced by ViDi Labs Pty.Ltd. 2010~2015 ABN P.O. Box 562, Mascot, NSW 1460, AUSTRALIA Prepared by V. Damjanovski 2015 Page 1

2 Please handle your Test Chart with care. The ViDi Labs SD/HD Test Chart was designed primarily for indoor use. If used outdoor, avoid direct exposure to rain, snow, dust, or long periods of exposure to direct sunlight. Although the ViDi Labs Test Chart has been designed specifically for the CCTV Industry, it can be used to verify the quality of other visual, transmission, encoding and recording systems. ViDi Labs Pty. Ltd. has designed this chart with the best intentions to offer an objective and independent measurement of various video signal characteristics, and although all the details are as accurate as we can make them, we do not take any responsibility for any damage or loss resulting from the use of the chart. This chart is copyrighted and cannot be copied or reproduced without a specific written permission. The chart design is subject to change without notice due to ongoing product improvements ViDi Labs Pty.Ltd. 2010~2015 A.B.N v. 10/04/2015 Prepared by V. Damjanovski 2015 Page 2

3 The SD/HD CCTV Test Chart In order to help you determine your camera resolution, as well as check other video details, ViDi Labs Pty. Ltd. has designed this special test chart in A3+ format, which combines three charts in one: for testing standard definition (SD) with 4:3 aspect ratio, high definition (HD) with 16:9 aspect ratio and mega pixel (MP) cameras and systems with 3:2 aspect ratio. We have tried to make it as accurate and informative as possible and although it can be used in the broadcast applications it should not be taken as a substitute for the various broadcast test charts. It s primary intention is to be used in the CCTV industry, as an objective guide in comparing different cameras, encoders, transmission, recording and decoding systems. Using our experience and knowledge from the previously designed CCTV Labs test chart, as well as the feedback we have had from the numerous users around the world, we have designed this SD/HD/MP test chart from ground up adding many new and useful features, but still tried to preserve the useful ones from the previous design. We kept the details used to verify face identification as per VBG (Verwaltungs-Berufsgenossenschaft) recommendation Installationshinweise für Optische Raumüberwachungs-anlagen (ORÜA) SP 9.7/5, and compliant with the Australia Standard AS With this chart you can check a lot of other details of an analogue or digital video signal, primarily the resolution, but also bandwidth, monitor linearity, gamma, color reproduction, impedance matching, reflection, encoders and decoders quality, compression levels, details quality in identifying faces, playing cards, numbers and characters, and last but not least - the pixel shifting effect due to moving objects. Prepared by V. Damjanovski 2015 Page 3

4 Before you start testing Lenses For the best picture quality you must first select a very good lens (that has equal or better resolution than the CCD/CMOS chip itself). In order to minimise opto-mechanical errors, typically found in vari-focal lenses, we suggest to use good quality fixed focal length manual iris lens, or perhaps a very good manual zoom lens. The lens should be suitable to the chip size used on the camera, i.e. it s projection circle should cover the imaging chip completely, in addition to offering superior resolution for the appropriate camera. Avoid vari-focal lenses, especially if resolution is to be tested. NOTE: If your lens has lower resolution then the camera you are testing, then you will have a false conclusion about the camera and the system you are testing. Shorter focal lengths, showing angles of view wider than 30, should usually be avoided because of the spherical image distortion they may introduce. A good choice for 1/2 CCD cameras would be an 8 mm, 12 mm, 16 mm, or 25 mm lens. For 1/3 CCD cameras a good choice would be when 6 mm, 8 mm, 12 mm or 16 mm lens is used. Since the introduction of megapixel and HD cameras there are mega-pixel lenses that can be used. Although the name mega-pixel on the lens may not necessarily be a guarantee for superior optical quality, one would assume this would be better quality than just an average vari-focal CCTV lens. Monitors Displaying the best and most accurate picture is of paramount importance during testing. If you are using the chart for testing only analogue SD cameras and systems, it is recommended that you use a high resolution interlaced scanning CRT monitor, with CVBS (Composite Video Signal) input and underscanning feature. High resolution in the analogue world means - a monitor that could offer at least 500TVL. Colour monitors are acceptable only if they are of broadcast, or near-broadcast, quality. Such monitors are harder to find these days, as many monitor manufacturers have ceased their production, but good quality CVBS monitors can still be found. A good try could be the supplier of broadcast equipment close to you. Understandably, cameras having over 500 TV lines of horizontal resolution cannot have their resolution tested with such monitors, but even higher quality are needed, only found in the broadcast industry. In the past, when testing camera resolution the best choice were high quality monochrome (B/W) monitor since their resolution, not having RGB mask, reaches 1000 TV lines. Unfortunately they are almost impossible to find these days. Screen shot of an actual HD camera output Prepared by V. Damjanovski 2015 Page 4

5 The next and more common option for good quality display nowdays are LCD and plasma screens. Some CCTV suppliers offer LCD monitors with BNC or RCA inputs for composite video signals too. If an LCD monitor is used for your analogue cameras testing, it is important to understand that the only way to have a composite video displayed on an LCD monitor is by way of converting the analogue signal to digital (A/D) by the monitor itself. The quality of such a conversion, combined with the quality of the LCD display, define how fine details you can see. The LCD monitor may have a high resolution (pixel count) which might be good for your computer image, but may fail when showing composite video. The reason for this could be the A/D conversion itself and the quality of this circuit (A/D precision and the upsampling quality). So, caution has to be excercised if using LCD monitors with CVBS input for measuring analogue camera resolution. When using LCD monitors for testing your digital SD, HD and MegaPixel cameras, the first rule is to use the video driver card (of the computer that decodes the video) to run in the native resolution mode of the monitor. The native resolution of the monitor is usually shown in the pixel count specification of the monitor. Furhermore, if an SD video signal is for example decoded and displayed on an LCD monitor with higher pixel count than the SD signal itself (e.g. PAL digitised signal of 768x576 pixel count is displayed on an XGA monitor of 1024x768) then the best image quality of the SD signal would be if it is shown in the native resolution of that image, e.g. 768x576. NOTE: The PAL digitisation as per ITU-601 produces a digital picture frame of 720x576 (also known as D1), but when this is decoded, there is a correction of the non-square pixels, so that 720x576 gets converted to 768x576 pixels with 4:3 aspect ratio. Similar applies to NTSC signals, where from 720x480, the non-linearity correction produces a 640x480 signal. Screen sizes and their Acronyms, based on their pixel count (courtesy of Wikipedia) Prepared by V. Damjanovski 2015 Page 5

6 Tripod If you are using longer focal length lens on your camera, this will force you to position the camera further away from the test chart. For this purpose it is recommended that you use a photographic tripod. Some users prefer to use vertical setup rather than horizontal, whereby the test chart is positioned on the floor and the camera up above looking down. This might be easier for the perpendicularity setup. In this case, a larger tripod is recommended with adjustable mounting head, so that a camera can be positioned vertically, looking at the test chart. There are photographic tripods on the market which are very suitable for such mounting. Light When the test chart is positioned for optimal viewing and testing, controlled and uniform light is needed for illuminating the test chart surface. One of the most difficult things to control is the colour temperature of the source of light used in the testing. This is even more critical if colour white balance is tested. Caution has to be exsercised if correct colour and white balance is tested as there are many parameters influencing this values. Traditionally, tungsten light are used to illuminate the chart from either sides, at a steep angle enough so as to not cause reflection from the test chart surface, but at the same time illuminate the chart uniformly. Tungsten light has different colour temperatures depending on the wattage and the type of light. Typically, a 100W tungsten light globe would produce an equivalent colour Prepared by V. Damjanovski 2015 Page 6

7 temperature of 2870 K, whilst a professional photographic lights are designed to have around 3200 K. Doing a resolution measurement as per broadcast standards requires illumination of 2000 lux at 3100 K light source. In CCTV, we allow for variation on these numbers since we rarely have controlled illumination as in broadcast, but it is good to know what are the broadcast standards. With the progress of lighting technology it is now possible to get solid state LEDs light globes with very uniform distribution of light (which is the important bit when checking camera response). In practice, many of you would probably use natural light, in which case the main consideration is to have uniform distribution of the light across the chart s surface. The chart is made of a matte finish in order to minimise reflections, but still care should be taken not to expose the chart to direct sunlight for prolonged periods of time as the UV rays may change the colour pigments. The overall reflectivity of the ViDi Labs test chart v.4.1 is 60%. This number can be used when making illumination calculations, especially at low light levels. Prepared by V. Damjanovski 2015 Page 7

8 Testing SD / HD or MP This test chart has actually three aspect ratios on one chart. The aspect ratios, as well as the resolution of each part has been accurately calculated and fine tuned so that the chart can be used as a Standard Definition test chart, with aspect ratio of 4:3, as a High Definition test chart with aspect ratio of 16:9 and as a MegaPixel with aspect ratio of 3:2. Since most of the measurements would be made with SD and HD cameras, we have made indicators for the SD to be white in colour (the edge triangles and focus stars), and the indicators for the HD to be yellow in colour (the edge triangles and the focus stars). Similar logic refers to the indicators of the SD analogue resolution in TVL (usually black text on white or gray) and the resolution in pixels for HD shown with yellow numbers (under the sweep pattern), or black on yellow area (for the resolution wedges). Only when analogue (SD) camera is adjusted to view exactly to the white/black edges the measurements for resolution, bandwidth, face identification and the other detail parameters will be accurate. Only when HD camera is adjusted to view exactly to the yellow/black edges - the measurements for resolution, pixels count, face identification and the other detailed parameters will be accurate. Finally, a MegaPixel camera with 3:2 aspect ratio can also be tested, and in this case the complete chart has to be in view, up to the white/black arrows top and bottom, and up to the yellow/black arrows left and right. In such a case, an approximate pixel count can be measured using the yellow/black numbers. NOTE: Since there are two HD standards, the SMPTE296 (usually referred to as HD720p), and the SMPTE 274 (usually referred to as HD1080 or True HD), in this test chart,we have inserted indicators for details for both of these standards, and they are depicted as HD720 and HD1080. Prepared by V. Damjanovski 2015 Page 8

9 Setup procedure Position the chart horizontally and perpendicular to the optical axis of the lens. NOTE: The accurate positioning of the camera for SD and HD systems respectively, refers to testing the resolution, pixel count, bandwidth, face identification, playing cards and number-plate detection. Other visual parameters, such as colour, linearity, A/D conversion and compression artefacts can be determined/measured without having to worry about the exact positioning. When testing SD systems - the camera has to see a full image of the SD chart exactly to the white triangles around the black frame. To see this you must switch the CVBS monitor to underscan position so you can see 100% of the image. Without having underscanning monitor it is not possible to test resolution accurately. When testing HD systems - the camera has to see a full image of the HD chart exactly to the yellow triangles/ arrows around the black frame. Illuminate the chart with two diffused lights on both sides, while trying to avoid light reflection off the chart. For more accurate resolution test, the illumination of the test chart, according to broadcast standards, should be around 2000 lux, but anything above 1000 lux may still provide satisfactoryt results, as long as this illumination is constant. Prepared by V. Damjanovski 2015 Page 9

10 It would be an advantage to have the illuminating lights controlled by a light dimmer, because then, you can also test the camera s minimum illumination. Naturally, if this needs to be tested, this whole operation would need to be conducted in a room without any additional light. Also, if you want to check the low light level performance of your camera you would need to obtain a precise lux-meter. When using colour cameras, please note that most cameras have better colours when switched on after the lights have been turned on, so that the colour white balance circuit detects its white point. Position the camera on a tripod, or a fixed bracket, at a distance which will allow you to see a sharp image of the full test chart. The best focus sharpness can be achieved by seeing the centre of the Focus target section. Set the lens iris to the middle position (F/5.6 or F/8) as this is the best optical resolution in most lenses and then adjust the light dimmer to get a full dynamic range video signal. In order to see this, an oscilloscope will be necessary for analogue cameras. For digital, or IP cameras, good quality computer with viewing/decoding software will be needed. Care should be taken about the network connection quality, such as the network cable, termination and the network switch. For analogue cameras, make sure that all the impedances are matched, i.e., the camera sees 75 Ohms at the end of the coaxial line. When measuring minimum illumination of a camera, it is expected that all video processing circuitry inside the camera electronics are turned off, e.g. AGC, Dynamic Range, IR Night Mode, CCD-iris, BLC and similar. Prepared by V. Damjanovski 2015 Page 10

11 What you can test To check the camera resolution you have to determine the point at which the five wedge lines converge into four or three. That is the point where the resolution limits can be read off the chart, but only when the camera view is positioned exactly to the previously discussed white/black or yellow/black arrows. The example on the right shows a horizontal resolution of approximately 1300 pixel when HD camera is tested. If, hypotetically, this image was from a 4:3 aspect ratio camera, it would have had an equivalent analogue resolution of approximately 900TVL. If you want to check the video bandwidth of the signal, read the megahertz number next to the finest group of lines where black and white lines are distinguishable. On the right example, one can notice that the analogue bandwidth indication ends up at 9.4MHz (or 750TVL) which is sufficient to cover what an analogue SD camera can produce. The real example to the right shows bluring of the 1400 pixels pattern. This is the same consistent camera result as in the example explaining the wedges above. Image detail of an actual HD camera output Image detail of an actual HD camera output NOTE: A camera that is designed to produce a True HD signal (1920x1080) doesn t neccessarily mean it will show 1920 pixels on the test chart. Details can be lost due to compression, misalignment, low light or bad lens/focus. The concentric star circle in the middle, as well as around the chart corners, can be used for easy focusing and/or backfocus adjustments. Prior to doing this, you should check the exact distance between the camera and the test chart. In most cases, the distance should be measured to the plane where the CCD chip resides. Some lenses though, may have the indicator of the distance referring to the front part of the lens. Prepared by V. Damjanovski 2015 Page 11

12 The main circle may indicate the non-linearity of (usually) a CRT monitor, but it can also be used to check A/D circuitry of the cameras, or monitor stretching, like in cases when there is no pixel for pixel mapping. The imaging CCD/CMOS chips, by design, have no geometrical distortions, but it is possible that A/D or compression circuitry introduces some non-linearity and this can be checked with the main circle in the middle. The big circle in the centre can also be used to see if a signal is interlaced or progressive (progressive would show smoother lines). The smaller starred circles around the corners of the chart can be used not only for focus and backfocus adjustments, but also for checking the lens distortions, which typically appears near the corners. On some cameras, it is possible to have lens optical axis misaligned, i.e. lens not being exactly perpendicular to the imaging chip, in which case the four small starred circles around the test chart will not appear equally sharp. The wide black and white bars on the right-hand side have twofold function. Firstly, they will show you if your impedances are matched properly or if you have signal reflection, i.e. if you have a spillage of the white into the black area (and the other way around), which is a sign of reflections from the end of the line of an analogue camera. The same clean black/white bars can show you the quality of a long cable run (when analogue cameras are tested), or, in the case of a DVR/ encoder/decoder - it s decoding/playback quality. The kids heads shots, as well as the white and yellow patterns on the righthand side, can be used to indicate face identification as per Australian Standard AS where a person s head needs to ocupy around 15% of the SD test chart height. This is equivalent to having 100% person s height in the filed of view, as per AS The equivalent head dimensions have been calculated and represented with another two smaller shots of the same, one referring to HD720 and the other to HD1080 when using the 16:9 portion of the chart. The same can be measured by using the white and yellow patterns, as per VBG (Verwaltungs- Berufsgenossenschaft) recommendations. The white pattern refers to SD cameras with 4:3 aspect ratio and the yellow ones refer to HD720 and HD1080 respectively. If you can distinguish the pattern near the green letter C, then you can identify a face with such system. If your system can further distinguish B, or even better A pattern, then the performance of such a system exceeds when compared to a system where only C can be distinguished. But, distinguishing the C pattern only is sufficient to comply with the standards. NOTE: It is the total system performance that define the measured details. This includes the lens optical quality and sharpness, the angle of coverage (lens focal length), the camera in general (imaging chip size, number of pixels, dynamic range, noise), the illumination of the chart, the compression quality, the decoder quality and finally the monitor itself. This is why this whole testing refers to system measurement rather than camera only. It is assumed that the observer has 20/20 vision. Prepared by V. Damjanovski 2015 Page 12

13 Furthermore, the skin colour of the three kids faces will give you a good indication of the cocasian flesh colours. If you are testing cameras for their colour balance you must consider the light source color temperature and the automatic white balance of the camera, if any. In such a case you should take into account the colour temperature of your light source, which, in the case of tungsten globes, is around 2800 K. Simplest and easiest to use is the daylight for such testing. Avoid testing colour performance of a camera under fluoro-lights, or mixed sources of light (e.g. tungsten and fluoro). The colour scale on the top of the chart is made to represent typical broadcast electronic colour bars consisting of white, yellow, cyan, green, magenta, red, blue and black colours. These colours are usually reproduced electronically with pure saturated combinations of red, green and blue, which is typically expressed with intensity of each of these primary colours from 0 to 255 (8-bit colours). Such co-ordinates are shown under each of the colours. If you have a vectorscope you can check the colour output on one of the lines scanning the colour bar. Like with any colour reproduction system, the colour temperature of the source is very important and in most cases it should be a daylight source. NOTE: The test chart is a hard copy of the computer created artwork. Since the test chart is on a printed medium, it uses different colour space then the computer colour space (subtractive versus additive colour mixing). Because of these differences - it is almost impossible to replicate 100% accurately these colours on paper. We have certainly used all available technology to make such colours as close as possible, by using Spyder3 colour matching system, but with time, and diffent exposure of the chart to UV and humidity, the colour pigment ages and changes. For these reasons we do not recommend using the colour bars as an absolute colour reference. The RGB continuous colour strip below the colour bars shown above, is made to have gradual change of colours, from red, through green and blue at the end. This can be used to check how good a digital circuitry, or how high compression, an encoder uses. If the end result shows obvious discountinuity in this gradual change of colours - it would indicate that either the encoder or the level of compression is not at its best. Similarly, using the gray-scales at the bottom of the chart, a few things can be checked and/ or adjusted. Using the 11 gray-scale steps Gamma response curve of camera/monitor can be checked. All 11 steps should be clearly distinguished. Monitors can also be adjusted using these steps by tweaking the contrast/brightnest so as to see all steps clearly. When doing so, analogue cameras have to be set so that the video signal is 1 Vpp video signal, while viewing the full image of the test chart. Observe and note the light conditions in the room while setting this up, as this dictates the contrast/brightness setting combination. The continuous changing strips, from black to white and from white to black, are made so that their peak of white, and their peak of black respectively, comes in the middle of the strips. These can also be used to verify and check A/D conversion of a streamer, encoder or compression circuitry. The smoother these strips appear when displayd on a screen the better the system. Prepared by V. Damjanovski 2015 Page 13

14 Always use minimum amount of light in the monitor room so that you can set the monitor brightness pot at the lowest position. When this is the case the sharpness of the electron beam of the monitor s CRT is maximum since it uses less electrons. The monitor picture is then, not only sharper, but the lifetime expectancy of the phosphor would be prolonged. Modern day displays (LCD, plasma and similar) do not use electronic beam which could be affected by the brightness/contrast settings, but they will also display better picture, and better dynamic range if the brightness/contrast are set correctly, using the 11 steps mentioned previously. Lately, there are an increasing number of LCD monitors with composite video inputs, designed to combine a CCTV analogue display, as well as HD. As noted in the very beginning of this manual, under the Monitor heading, please be aware of the re-sampling such monitors perform in order to fill-up a composite analogue video into a (typically) XGA screen (1024X768 pixels). Because of this, LCD monitors are not recommended for resolution testing. If image testing needs to be done using a frame grabber board on a PC use the highest possible resolution you can find, but not less than the full ITU-601 recommendation (720X576 for PAL, and 720X480 for NTSC). Again, in such a case, native camera resolution testing can not be performed accurately as signal is digitised by the frame grabber. If however, various digital video recorders are to be compared then the artificial (digitised) resolution can be checked and compared. The ABC fonts are designed to go from 60 points font size for A down to 4 points for Z. This could also be used for some testing and comparisons amongst systems. For the casino users, we have inserted playing cards as visual references. The cards may be used to see how well your system can see card details. If you can recognise all four of the cards (the Ace, the King, the Queen and the Jack) then your system is pretty good. Similar logic was used with the playing card setup for SD, HD720 and HD1080 standards, hence there are three different cards sizes. It goes without saying that when the SD cards are viewed the CCTV camera should be set to view the 4:3 portion of the chart, exactly to the white/black arrows. Similarly, when the HD cards are viewed, the camera has to be set to view the 16:9 portion of the test chart, exactly to the yellow/black arrows. Typically, playing cards height should not be smaller than approximately 50 pixels on the screen, irrespective of whether this is an SD, HD720 or HD1080 system. The playing cards different sizes in this test chart are calculated so that they are displayed on your screen at approximately 50 pixels height. Prepared by V. Damjanovski 2015 Page 14

15 NOTE: In casino systems the illumination levels are very low, typically around 10 lux or lower. Such a low illumination may influence cards recognition too, so if realistic testing is to be done, the chart illumination should be around the same low levels. Finally, the four corners of the test chart have 90% black and 10 white areas. Although these corners fall outside the 4:3 and the 16:9 chart areas, they may still be used to check on system reproduction quality. If you can distinguish these areas from the 100% black border or 100% white (0% black) frame with the 3:2 numbers, it suggests your system overall performance is keeping such details and can be classified as very good. If this is not the case, adjustments need to be done either in the camera A/D section (typically where Gamma or brightness/contrast settings are) or where encoder/ compression section is. The below images are of an actual HD camera testing with various lenses and at various light levels. On the last page there is a graphical representation of all the points of interest on this test chart and what their applications are. IMPORTANT NOTE: The lifetime expectancy of the colour pigments on the test chart is approximately 2 years, and it can be shorter if exposed longer periods to sunlight or humidity. It is therefore advised that a new copy is ordered after 2 years. A special discount is available for existing customers. Prepared by V. Damjanovski 2015 Page 15

16 Testing the effect of moving objects Very often, when objects are moving in front of a camera with certain speed (velocity) they will appear blurry when playing back, and a frame is paused. Some refer to this blurinness as object s pixel shifting during the exposure. This pixel shifting occurs due to objects moving more than a pixel length during the electronic exposure of the camera sensor (typically 1/25s or 1/30s for live viewing). The faster the speed of movement, the more obvious the blurriness will be. There are other factors that influence the appearance of blurriness. These are the lens angle of view, the size of the sensor (pixels), and the distance of the object(s) to the camera. Using the ViDi Labs SD/HD test chart it is possible to test and measure the effect of fast moving objects on the resulting recorded video. In this chapter we have advised this process, and offered some formulas that will help in determining such pixel shifting, while taking into account all the possible variables. The formula below summarises all these parameters into a single expression where the resulting value is the number of pixels shifts. p V = (f v t) / (d p s ) where: p V = pixels shift due to movement f = focal length of the lens in use (mm) v = velocity (speed) of movement in m/s t = exposure time in seconds (e.g. 1/25s = 0.25s) d = distance to the moving object p s = pixel size in mm (e.g. 2.2um = mm) In order to test and evaluate the movement effect with the ViDi Labs SD/HD test chart, a small modification of the test chart needs to be performed. You will need a small Philips screwdriver, of 1mm in diameter and about 3m length of fine, but strong sewing string. Fishing line can also be used. Puncture small holes on all four corners of the test chart, around 3mm from the edges. Then, with a sewing needle, or using the small screwdriver, push the string through the holes, so that it makes a symetrical long hanger for the test chart (as in the picture). The test chart can now be hanged on a smooth wall, using (preferably) a screw intended for wood or plaster boards. The aim is to make the test chart be able to oscillate like a pendulum when pushed on one side and released. It is the amount of movement and the appearance of the test chart on the recorded footage that will be used to evaluate and quantify the motion effect. Prepared by V. Damjanovski 2015 Page 16

17 When the string is put through the four corners of the test chart it should make a pendulum with the test chart hanging down around 1m away from the hanging point. Using a pencil, mark the corners of the test chart when it is hanging down, on the wall. These markers can later on be used as a reference. Position and focus a camera to view approximately 1m wide, i.e. just about over twice the width of the chart. This way, the camera should have wide enough view to see the chart oscillating, but also close enough view to be able to see specific positions marked during the oscillatory movement of the chart. These positions can later on be used to calculate the velocity of movement in cm/s, which can then be converted into m/s. Move the test chart with one finger to the left, up to the point where the top right hand corner of the chart aligns with the marker for the top left corner. This way, we have displaced the test chart for around 416mm, which is the width of the test chart. Then, let the test chart go and oscillate. Because the chart is very light and the pendulum length is long, the test chart will oscillate for a couple of minutes until it settles completely. During this time, let the camera view the chart and record the oscillation. Then, play back the recorded footage. If you recording software counts seconds and miliseconds, it should be relatively easy to measure the traversing speed of the oscillation. Using markers you should make with a pencil, it will be relatively easy to Prepared by V. Damjanovski 2015 Page 17

18 calculate the speed of oscillating. Basically, all you have to do is measure how many cm is the chart moving during a second, while slowing down playback, or even better, playing frame by frame until 1 second passes. Although this oscillation is a part of an arc and it is better described as a part of revolving motion, the pendulum radius is long enough to be approximated as a linear motion.at it s lowest point (when the chart passes through it s horizontal position). As the formula shows, the variables that need to be entered are the focal length of the lens, the speed (velocity) of movement (which is calculated from the recorded footage), the exposure time (typically 0.25s or 030s, but can be used for any other exposure, including the long or shorter than live ), the distance to the object and finally the pixel size. The pixel size is easily calculated by dividing the sensor width (in mm) with the number of horizontal pixels. Example: HD camera, with 4.2mm sensor width (s), producing live 1080HD video (hence p s = 4.2/1920 = mm) Normal live exposure of t = 1/25s = 0.04 s and Distance from camera to scene is 1.2 m. The Lens focal length = 8mm and the measured velocity of the moving object = 0.16m/s (0.6km/hr). The Calculated pixel shift p V = (8 x 0.16 x 0.04) / (1.2 x ) = / = 19 pixels Prepared by V. Damjanovski 2015 Page 18

19 Some real-life examples Here are some more snap-shots from various real camera testing. The quality of this reproduction is somewhat reduced by the PDF/JPG compression of images, as well as the quality of this print in the book, but the main idea can still be seen. For example, the four snippets below are from a same HD camera, using H.264 video compression, set to 4 Mb/s, using four different lenses, all of which are claimed to be "mega-pixel" lenses. The snippets shown below are colour and HD resolution (1920 pixels wide), but this book print is in B/W so the readers may not see the real difference, but anybody interested in it can me and I wil make the actual files available for inspection. It is quite obvious that the first lens has the worst optical quality, and the last has the best. If you would to use this camera/lens combination for face identification, the first one would hardly qualify. The smallest pattern in the yellow group of patterns is not distinguishable, which means face identification will hardly happen. Yet, as shown in the bottom snippet, the same camera with the same compression settings and another (better) lens will pass. Four different "mega-pixel" lenses produce different results on the same sensor using the same compression settings Prepared by V. Damjanovski 2015 Page 19

20 Another real example snapshots are shown below, where two different lenses are tested and compared. Namely, of the two lenses, the top one obviously looks sharp throughout the test chart, while the bottom one appears reasonably sharp in the middle, but becomes blurry as you go towards the periphery of the test chart. Lens with non-uniform sharpness (bottom) A test chart at 1.1 lux illumination is below. Resolution in low light can not be measured due to high noise content. Resolution is dramatically reduced at low light levels Prepared by V. Damjanovski 2015 Page 20

21 As indicated below, lenses viewing the test chart from a very close distance usually produce geometric barrel distortion. It is not recommended for accurate resolution measurement, but, if there is no choice, it may still allow for reasonably good quality measurement Short focal length lenses produce barrel distortion, not suitable for resolution test The ViDi Labs test chart can be used for other various testing too, precise focusing at certain distance for example, or just simply to compare details and video reproduction in various circumstances. The example on the right shows just one more way of using the ViDi Labs test chart for checking dynamic range of various cameras, as conducted by IPVM. And last, but not least, on the next page there is a graphical summary of the key measuring points in the test chart and their meaning. Dynamic range testing by IPVM Prepared by V. Damjanovski 2015 Page 21

22 Prepared by V. Damjanovski 2015 Page 22